The Research Of PMSM Motor Control Strategy Based On Deep Learning

Due to the high-power factor,high starting torque,short starting time and high overload capability,the permanent magnet synchronous motor is widely used in the fields of robot arms,robots,electric vehicles,etc.Permanent magnet synchronous motor vector control technology and direct torque control technology have been developed rapidly since its introduction.However,vector control requires complex rotor rotation coordinate system transformation calculation and sensitivity to motor parameters,resulting in a large number of harmonics in the actual motor torque;traditional direct torque control technology uses two Bang-Bang controllers for torque and magnetic respectively,and the response is rapid,but there are also large flux linkages and torque ripples.Because of the insensitivity of neural network technology to parameters and the strong nonlinear fitting ability of neural networks,deep learning neural networks can be used to replace the complex mathematical transformations in vector control technology or the torque and magnetic in direct torque control,thereby achieving the effect of reducing the torque ripple.In this thesis,the permanent magnet synchronous motor is taken as the control object,and the simulation research and physical research of the motor control strategy based on deep learning are carried out.In this thesis,a novel permanent magnet synchronous motor vector control system is designed based on DNN neural network.The simulation control model of permanent magnet synchronous motor based on DNN neural network is designed by MATLAB2017b/Simulink.The simulation results show that the permanent magnet synchronous motor control system based on DNN neural network effectively reduces the torque ripple harmonics of about 95% of the motor simulation operation.The torque ripple coefficient is reduced from 0.15 of the traditional PI vector control to 0.0075.At the same time,the stator three-phase current waveform output from the model is improved.Then,the hardware control platform based on SGMJV-08AAA61 permanent magnet synchronous motor is built.The hardware circuit of motor drive,control and the hardware circuit of current,temperature and speed acquisition are designed.The corresponding PCB board is drawn and manually welded.Finally,the debugging is successful.The platform was used to test the torque fluctuation suppression effect.The experimental results show that the permanent magnet synchronous motor control system based on DNN neural network effectively reduces the motor torque harmonic of about 83%.The torque ripple coefficient is reduced from 0.12 of the traditional PI vector control to 0.015,which is feasible for engineering applications.Finally,a direct torque control system for permanent magnet synchronous motor based on DNN neural network is proposed and related simulation research is done.The simulation results show that the direct torque control system based on DNN neural network has achieved some improvement over the traditional direct torque control system in terms of electromagnetic torque fluctuation,rotational speed fluctuation and stator flux linkage fluctuation.